[PW_Forum] Ultrasoft Troubles

[John B Burdette]

Dear users,

     I am attempting to generate a good ultrasoft 
pseudopotential for Tungsten (Z=74).  My problem lies in 
making sure if the PP is good, I ensure that the logarithmic 
derivatives match those of the all-electron run through a 
graphing program called "xm-grace".  I generally just look to 
see if the two logarithmic derivatives agree with each other, 
and more often than not they do, but this time I'm having 
troubles with the 6S orbital for Tungsten.  The graph shows a 
difference in energies at about 0.08 Ryd, but here is the 
output of the ld1.x program for my input file.

n l     nl             e AE (Ryd)       e PS (Ryd)   De AE-PS 
(Ryd) 
4 3     4F   1(13.00)       -3.12165       -3.12235        
0.00070
3 2     5D   1( 5.00)       -0.41851       -0.41864        
0.00012
1 0     6S   1( 2.00)       -0.44416       -0.44416        
0.00000

So as you can see, the output tells me that the difference in 
energies is very small, but what I view is relatively large 
in comparison.  Any clue as to why this happens or what I can 
do to fix it?

Best Regards,

John